Easier to disambiguate in

[jishnub]

In this PR, instead of specializing in(v, ::T) where {T<:TypedEndpointsInterval}, we define

 in(v, I::TypedEndpointsInterval) = _in(v, I)

and specialize _in. This makes it easier for other packages to define in(v::CustomType, I::TypedEndpointsInterval) without having to resolve ambiguities with more specific types.

[codecov]

Codecov Report

Base: 99.16% // Head: 99.17% // Increases project coverage by +0.00% 🎉

Coverage data is based on head (e979412) compared to base (0661291).
Patch coverage: 100.00% of modified lines in pull request are covered.

❗ Current head e979412 differs from pull request most recent head ddf5679. Consider uploading reports for the commit ddf5679 to get more accurate results

Additional details and impacted files

@@           Coverage Diff           @@
##           master     #126   +/-   ##
=======================================
  Coverage   99.16%   99.17%           
=======================================
  Files           3        3           
  Lines         239      241    +2     
=======================================
+ Hits          237      239    +2     
  Misses          2        2           

Impacted Files	Coverage Δ
src/IntervalSets.jl	98.33% <100.00%> (+0.02%)	⬆️

Help us with your feedback. Take ten seconds to tell us how you rate us. Have a feature suggestion? Share it here.

☔ View full report at Codecov.
📢 Do you have feedback about the report comment? Let us know in this issue.

[hyrodium]

LGTM! Could you bumb the version to v0.7.4?
(x-ref: 0661291#commitcomment-91522252)

[jishnub]

Should we instead define in(v, t::Domain) = _in(v, t), with the fallback

_in(v, t::Domain) = throw(ArgumentError("in(::$(typeof(v)), ::$(typeof(t))) is not defined"))

TypedEndpointsInterval seems like an internal type that packages might not be willing to disambiguate against.

EDIT: DomainSets.jl seems to be committing type-piracy with this currently, so it would be good to move those methods here

I've instead defined in(v, t::AbstractInterval) here.

Edit 2: Defining in(v, t::AbstractInterval) will actually create a more specific method than in(v, ::Domain) that is currently defined through piracy in DomainSets. This might lead to breakages. I've reverted to defining in(v, ::Domain) here. This means one would obtain "method overwritten" warnings when loading DomainSets, but nothing will break. We should eventually move the methods from DomainSets to this package, or perhaps to DomainSetsCore

Edit 3: This seems harder than I had thought because of the type-piracy. I've gone all the way back to the original idea presented in the first post.

[hyrodium]

I agree with all of your points. It seems better to have DomainSetsCore.jl package. (x-ref: #106 (comment))